Rearrangements of 2-Bicyclo[3.2.0]heptyl Grignard Reagents ratio of 25/24 was 17/83 after 15 min; the ratio did not change after an additional 40 min at -100 "C. An additional 1.1 equiv of n-butyllithium was added. After 15 min the above ratio was 75/16; compound 26 w,ts also detected. The mixture was continually stirred at -100 "C; examination of aliquots showed that the amount of 24 decreased while the amount of butylated products (26 and 27) increased. The mixture was quenched with water after a total of 4 h after the second addition of n-CaHgLi. The mixture of products obtained contained phenylpropionitrile (25, 62%), abutyl-0-phenylpropionitrile (26, 19%), a-butyl-P-(p-butylpheny1)propionitrile (27, 13%), and an unidentified product. Products were collected by preparative GLC. a-lButyl-j3-phenylpropionitrile(26): NMR (CDC13) 6 0.9 (t, 3, aliphatic H), 1.55 (m, 6, aliphatic H), 2.9 (m, 3, aliphatic H), 7.3 (m, 5,aromatic H). Anal. Calcd for C13H17N: C, 83.37,H, 9.15;N, 7.48. Found: C, 83.55;H, 9.05;N, 7.55. a-Butyl-8-(p-butylpheny1)propionitrile(27): NMR (CDCl3) 6 0.70-1.9(m, 16,aliphatic H), 2.2 (m, 2,CHz), 2.85 (m, 3,aliphatic H), 7.2 (m, 4,aromatic H). Anal. Calcd for C17H25N: C, 83.89;H, 10.35.Found: C, 83.79;H, 10.60. p-Iodobenzylnitrile (28). E ~ a m i n a t i o n ~of~ an aliquot, quenched with water taken after 15 min from reaction of 2tIz7with 1 equiv of n-C4HgLi at -100 "C, showed the ratio of benzylnitrile to starting material (28) to be 80/20;starting material immediately disappeared upon addition of an additional 0.5 equiv of n-C4HgLi. Attempts to trap the anionic products from the reaction mixture with cyclohexanone gave a multicomponent mixture (GLC) which was not resolved. Registry No.--1, 6952-59-6; 2, 100-47-0;3, 57775-02-7; 4, 613662-5;5, 623-00-7;6, 57808-43-2;7, 1503-49-7;8, 2042-37-7;9, 57775-03-8;10, 28873-85-0;11, 57775-04-9;12, 34446-14-5;15, 1009-14-9;16, 57775-05-0;17a, 19472-74-3;17b, 31938-07,5;17c, 16532-79-9;20, 57775-06-1;23, 57775-07-2;24, 57775-08-3;26, 54321-42-5;27, 57775-09-4; 28, 51628-12-7;n-butyllithium, 10972-8;benzophenone, 119-61-9;methyl benzoate, 93-58-3;cyclohexanone, 108-94-1;methyl acrylate, 96-33-3;phenyl isocyanate, 103-71-9;phenylacetonitrile, 140-29-4; n-butyl bromide, 109-65-9; a-butylphenylacetonitrile, 3508-98-3;ap-dibutylphenylacetonitrile, '3508-99-4;2-(o-bromophenyl)propion~trile, 57775-10-7;P(p-bromophenyl)propionamide, 57775-11-8; P-(p-bromopheny1)propionic acid, 1643-30-7.
J. Org. Chem., Vol. 41, No. 7, 1976
1191
References and Notes ( 1 ) Supported by US. Army Research Office, Grant DAHC04 74 GD128. (2) H. Gilman and G. Melstrom, J. Am. Chem. SOC., 70, 4177 (1948). (3) (a) W. E. Parham and Y. A. Sayed, J. Org. Chem., 39, 2051 (1974); (b) ibid., 39, 2053 (1974). (4) The degree of metalation was determined by treating aliquots with water and analyzing the dried organic products by (a) GLC [20% SE-30 on Chromosorb W (sol80 mesh), 6 ft X 0.25 in., or 5% SE-30 on Chromosorb W (60/80 mesh), 3 ft X 0.25 in., including injection of authentic starting halides and reduced products], (b) NMR, the benzylic protons of piodobenzyl nitrile and benzyl nitrile appear as sharp singlets at 6 3.75 and 3.65, respectively. (5) Dr. Robert Piccirilli, Duke University, private communication. (6)Thin layer chromatography showed it to be multicomponent. (7) Imino anthraquinones (or Kbutyl derivatives) were anticipated as possible products of self-condensation of 13 by analogy to products of selfcondensation of lithium dithiobenzoates (cf. ref 3b). (8) Reaction of 18a, formed from 17a and 1 equiv of n-butyllithium. with methyl iodide, gave a 79% yield of nearly pure 2-(o-bromophenyl)propionitrile (see Experimental Section). (9) Analysis of the organic products obtained by decomposition of aliquots with water showed 17% halogen-metal exchange after 0.5 h with 1 equiv of n-butyllithium. ( I O ) Butylation is assumed to occur by reaction of n-butyl bromide, formed by exchange, with the dilithio derivative formed from 24. ( 1 1 ) Studies of aliquots showed that 80% iodine-lithium exchange occurs when 28 is treated with 1 equiv of n-butyllithium. (12) 0-,m-, and pbromobenzonitrile are commercially available. (13) Practical grade stored over molecular sieves. (14) G. W. A. Kahlbaum and G. von Wirkner, Ber., 27, 1894 (1894). (15) (a) bp 90-110 ' C ; (b) bp 60-90 O C ; (c) bp 30-60 OC. (16) L. Novak and M. Protiva, Collect. Czech. Chem. Common., 24, 3966
(1959). (17) (18) (19) (20) (21)
F. Ahrens, Ber., 20, 2957 (1887). W. H. Puterbaugh and C. R. Hauser, J. Org. Chem., 29, 853 (1964). Freshly distilled from hydroquinone. M. P. H. van der Meulen, Reci. Trav. Chim. Pays-Bas, 15, 282 (1896). (a) C. L. Jackson and J. F. White, Am. Chem. J., 2, 315 (1880); (b) N. Campbell and J. E. McKail, J. Chem. SOC.,1251 (1948). (22) R. C. Fuson et al., J. Org. Chem., 23, 1161 (1958). (23) (a) W. E. Parham and D. C. Egberg, J. Org. Chem., 37, 1545 (1972); (b) L. Fieser and M. Fieser, "Reagents for Organic Synthesis", Voi. I, Wiley, New York, N.Y., 1967, p 450. (24) Shown by coinjection of an authentic sample prepared from benzyl nitrile ( 1 equiv), sodium hydride (2 equiv), and methyl iodide (2 equiv) in ~ (lit.25bp 232 dirnethylformamide: bp 226 OC (754 mm); n Z 51.5016 O C ; m 1.50665). (25) 0. Waliach, Chem. Zentralbl., 1047 (1899). (26) (a) S. Gabriel and 2. Zimmerman, Ber., 13, 1683 (1880); (b) J. V. Braun and J. Nelles, Chem Ber., 66, 1464 (1933). (27) (a) C. F. Maybery and C. L. Jackson, Am. Chem. J., 2, 253 (1880); (b) C. L. Jackson, ibid., 1, 100 (1879).
Ring Cleavage Rearrangements of 2-Bicyclo[3.2.0]heptyl and Related Grignard Reagents E. Alexander Hill,* Robert J. Theissen,la Charles E. Cannon,lb Richard Miller, Richard B. Guthrie, and Augustin T. ChenlC Department of Chemistry, University of Wisconsin-Milwaukee, Milwaukee, Wisconsin 53201 Receiued July 1, 1975
The Grignard reagent (12) from 2-bromobicyclo[3.2.0]heptane undergoes a ring cleavage rearrangement to cyclopentenylethyl (13) and cycloheptenyl (14)Grignard reagents. Grignard 14 is slowly converted to 13. The rate of rearrangement of 12 is thought to be somewhat retarded by geometric restrictions introduced by the bicyclic skeleton. The facility of the rearrangement of 12 14 may indicate that the preferred transition state for rearrangeoccurs in ment is nonplanar. Rearrangement of the Grignard reagent 25 from 3-chlorotricyclo[5.3.0.02~6]decane analogous fashion. Grignard reagents 31 from 2-bromo-6-alkoxybicyclo[3.2.0]heptanesdecompose with elimination of the alkoxy group and ring cleavage to 3-vinylcyclopentene and 1,4-cycloheptadiene. 4
T h e rearrangement of appropriate organomagnesium compounds, either by cleavage of a strained ring or its reverse (intramolecular addition to a multiple bond), is well established (eq l)?
a-
I
c---C--MgX
1
I
-
=C,?(
,CqMgX
/
Experimental results have been interpreted as being most consistent with a synchronous four-center process for this rearrangement.2-4 In t h e present paper, we report Grignard cleavage studies in the bicyclo[3.2.0]heptyl and tricy~ l o [ 5 . 3 . 0 . 0 ~ ~ ~ ] dsystems. ecyl These studies were undertaken to probe the effects of geometric constraints imposed on t h e transition state by the bicyclic system, and t o assess t h e influence of a polar substituent, a n alkoxy group. Preparation of Halides. Bicycl0[3.2.0]heptan-2-01 (1)
1192 J . Org. Chem., Vol. 41, No. 7, 1976
Hill, Theissen, Cannon, Miller, Guthrie, and Chen
was synthesized as reported by Winstein and S t a f f ~ r dRe.~ action of the alcohol with thionyl chloride in pyridine yielded a mixture containing exo- 2-chlorobicyclo[3.2.0]heptane and 7-chloronorbornane in a ratio of about 1:2. Bromide formation with bromine and triphenylphosphine was more successful in producing product without rearrangement. endo-2-Bromobicyclo[3.2.0]heptane,exo2-bromobicyclo[3.2.0]heptane, and 7-bromonorbornane were formed in a ratio of 5:2:1, respectively. The mixture could be resolved by gas chromatography, b u t was used without separation for t h e Grignard studies.
0
A ' ; "
-
L o 4
0
LiAIH,
7
9
8
2a, X = Br b, X = C1
1
5-Bromocycloheptene (3) was synthesized from the corresponding alcohol.6 I t also contained a small amount (-10%) of 7-bromonorbornane, undoubtedly produced via carbonium ion rearrangement in the reaction of the alcohol with phosphorus tribromide. The photochemical cycloaddition of cyclopentenone t o methyl or ethyl vinyl ether produced the corresponding 6alkoxybicyclo[3.2.0]heptan-2-ones 4, which were converted t o the respective bromides 6 (eq 2). Each of the compounds
Stereochemical Considerations. From the results of x-ray crystallographic8 and electron diffractiong studies, i t appears t h a t there is a pronounced tendency toward twisting or folding of the five-membered ring of the bicycloheptyl and tricyclodecyl systems. Folding of the four-membered ring is also likely. The precise manner of distortion from planarity may depend upon substitution and unsaturation present, as well as crystal forces in the x-ray studies. NMR spectra provide further evidence for nonplanar geometry, and also for the configuration of substituents. In the 2-substituted bicyclo[3.2.0]heptanes, H-2n generally appears as a doublet,1° with Jzn-sn as the only coupling of significant magnitude (see 10). In the present work, dou-
2x%L 2n
6X
0
10
II
CHOR
OR 4a,R = CH, b, R = CZH,
OR
5
OR 6
4-6 was obtained as a mixture of two isomers present in nearly equal amounts. These are epimers differing in configuration a t the alkoxy group, rather than a mixture of 6and 7-alkoxy derivatives, since either Wolff-Kishner reduction of ketones 4a or 4b or tri-n-butyltin hydride reduction of 6a led to a mixture of saturated bicyclic ethers, rather than a single ether product. Hydrolysis of the Grignard reagent from 6a or 6b, or reduction of 6a with sodium in alcohol, led t o a mixture which contained unsaturated components also, as a result of disproportionation and/or elimination. T h e epimers of 4-6 were separated on a small scale for spectroscopic characterization, but most of the work was done with the mixture. Attempted preparation of the chloride corresponding t o 6b (thionyl chloridepyridine) led to a mixture of products, including one lacking the ethoxy group, and was not pursued further. Tricyclo[5.3.0.02~6]decan-3-one (7), prepared by photocycloaddition of cyclopentenone t o ~ y c l o p e n t e n e ,was ~ reduced and converted t o the chloride 9 (eq 3).
-
blet absorption was noted for alcohol 1 and halides exo-2a, 3-4 Hz). Small values for J z n - l and J z ~ are- ~ ~ found also in unsaturated analogues.ll These require a conformation in which carbon 2 is rotated in the exo direction and carbon 3 endo, in order to obtain the implied dihedral angles of about 90'. With an endo 2 substituent, coupling of H-2x is more extensive. In the present work, spectra of alcohols 5 and 8 showed absorption resembling a broadened first order (1-3-3-1) quartet, suggesting coupling t o three hydrogens (1, 3x, 3n) with similar coupling constants of 7-8 Hz. Similar results have been reported for some more heavily substituted or unsaturated analogues,10bB11,12 though a different interpretation has also been given.loa The configurations of compounds in this study are consistent with chemical expectations, as well as with the NMR spectra. Thus, reaction of 1 with thionyl chloride gives largely carbonium ion rearrangement to 7-chloronorbornane; most of the product of unrearranged structure appears t o be exo isomer, formed with retention of configuration. Reaction with triphenylphosphine dibromide gives much less rearrangement, and most of the unrearranged product is now formed with inversion (endo), as expected from the greater tendency of this reagent t o react by direct displacement. Reaction of 1 with triphenylphosphine-carbon tetrachloride also gives unrearranged endo isomer, but much more rearrangement.loa In the preparative sequence of eq 2 and 3, hydride reduction of ketones occurs as expected from the exo direction t o yield endo alcohol. Conversion to halides without rearrangement was much more successful, since displacement by halide ion can occur from t h e less hindered exo side of the ring, and the C-C bond
6, and 9 (J
J . Org. Chem., Vol. 41, No. 7, 1976 1193
Rearrangements of 2-Bicyclo[3.2.0]heptyl Grignard Reagents
Table I. Rearrangement Rates of Bicyclo[3.2.O]heptyl and Related Grignard Reagents in Ether Grignard rearrangement
Concn, M
12 --c 13
0.16
Temp, O
0.65 0.16
14 -- 13
0.16 -1
-.
ero-3lb 32 -k 33 endo-3la --r 32 + 33 25 -+ 26 16a 1Gb
0.4 0.4 -1
0.145 0.06
krei (100 “C)
a
4.68 f 0.09 29.9 f 0.7 104 & 4 1.9 16 53 -50 0.45 2.4 7.3
87.6 102.7 114.6 79.6 91.8 106.6 112 87.6 102.7 114.6 114.6 110 125 80 76 110 66-80 66-94
0.23
12 -- 14
IO6 k , s-l
C
0.059
5 x 10-3
c
-10-3
-2
-0.3 -3.5 -80
-2 >40
>SO0 -1
-10-3 1.6ad (1.00)d.e
a Error limits vhere given are standard deviations from a least-squares correlation. Errors in other rates are probably about &lo%, unless otherwise indicated as approximate. AH: = 31.6 f 0.9 kcal/mol; AS* = 4.3 f 1.5 eu. AH*= 28 kcal/mol, AS* = -10 eu. Reference 17. e AHl = 31.9 4:0.7 kcal/mol, AS1 = 11 f 1 eu.
t h a t would have t o be involved in a rearrangement t o a norbornyl skeleton is no longer situated anti to the departing OH function. In NMR spectra of Grignard reagents prepared in this study, H-2 gave a poorly resolved multiplet, probably consistent with the anticipated mixture of exo and endo configurations of the magnesium halide function. The configuration a t c6 (or is less satisfactorily estqblished on the basis of coupling patterns of H-6n and H-6x. There are differences in coupling constants apparent in the spectra of unsaturated and more highly substituted compound^,^^ but the multiplets for compounds in eq 2 are complex and poorly resolved, precluding any really definitive distinction from coupling patterns. On the other hand, there does appear t o be a consistent chemical 4hift difference, with either a hydrogen or methyl appearing a t higher field when in the endo position.13J4 Interestingly, exo and endo methoxy groups appear t o have the same shift.15 In the present work, the isomers with endo methoxy or ethoxy groups had shorter GC retention times. T h e methoxy derivatives had more easily interpreted spectra. The conformation of cycloheptenes has been discussed by a number of workers.I6 Although there is some disagreement, it appears t h a t both chair (lla)and boat ( l l b ) forms may be reasonably accessible energetically. T h e boat form may undergo pseudorotation through the apparently high-
c,)
lla
llb
and acetate, is deshielded in the bromide so as t o fall within the multiplet of other ring hydrogens. In the Grignard reagent, the Cg proton is a pentuplet resulting from apparently equal coupling to four protons ( J 5.5 Hz). The difference could be attributed either to conformational difference, or to a “deceptively simple” situation. Unfortunately, the coupling patterns do not allow a clear choice among conformations. Examination of models suggests t h a t a n axial C5 hydrogen in either the chair or boat conformation should have one large and one quite small coupling constant, while a n equatorial Cg hydrogen in either conformation should have two rather small coupling constants. P r e p a r a t i o n a n d R e a r r a n g e m e n t of G r i g n a r d Reagents. Grignard reagents, prepared in ether or T H F solution from halides 2 , 3 , 6 , and 9, were heated in sealed tubes. Products of reaction were determined by hydrolysis of the reagent, followed by their separation with gas chromatography. In some instances, NMR spectra of the intact Grignard reagent provided evidence of the reaction. By analyzing products as a function of time, a measure of the rate of reaction could be obtained. Kinetics results are listed in Table I. This table also includes rough rate estimates made by analysis of only one t o three samples of heated Grignard rather than the minimum of six to eight tubes used for the better kinetics runs. The 2-Bicyclo[3.2.0]heptyl G r i g n a r d . Rearrangement of the unsubstituted 2-bicyclo[3.2.0]heptyl Grignard reagent 12 in ether solution was studied at temperatures between 80 and 115 OC (eq 4). Cleavage to the primary cyclo-
-
llc
er energy twist-boat ( l l c ) ,but the chair form is more rigid. N M R spectra of the 5-substituted cycloheptenes prepared in the present work are of some interest. The Cg proton of both the alcohol and its acetate appear as well-resolved multiplets, showing coupling t o two sets of two. protons, probably those cis and trans a t C4 and c6 (apparent J = 3.8, 8.6 Hz in the alcohol; 4.0, 8.0 Hz in t h e acetate). However, the bromide has only an unresolved multiplet of comparable width. T h e difference is probably a consequence of virtual coupling t o other ring hydrogens, rather than a difference in conformation; a two-proton multiplet (probably a t C4 and Ce), which appears a t high field in the alcohol
12
\
MgX
l3
(4)
14
pentenylethyl Grignard 13 as the major pathway was indicated by the appearance of a high-field CHzMg triplet in the NMR spectrum of the heated Grignard. Gas chromatographic analysis of the hydrocarbons formed by hydrolysis showed t h a t the secondary cycloheptenyl Grignard 14 was a minor rearrangement product (6.7% a t 114.6 OC t o 9.6% a t
Hill, Theissen, Cannon, Miller, Guthrie, and Chen
1194 J . Org. Chem., Vol. 41, No. 7, 1976 87.6 “C). At much longer heating times, it appeared t h a t 14 was slowly converted to 13. This was confirmed by independent preparation and rearrangement of 14. The conversion of 14 to 13 presumably occurred via cyclization t o 12, although this intermediate did not reach sufficient concentration t o allow the detection of any bicyclo[3.2.0]heptane on hydrolysis. Norbornane, resulting from minor amounts of 7-bromonorbornane present in both starting bromides, comprised a constant fraction of the hydrolysis products. In a previous publication,17 it was pointed out t h a t 2-bicyclo[2.l.l]hexyl Grignard reagents 15 undergo ring cleavage reactions considerably more slowly than the monocyclic analogue 16, despite a somewhat greater relief of ring strain
15
CH, 16a, X = c1
b, X
=
Br
during cleavage of the bicyclic organomagnesium. The difference was attributed to the inability of the more rigid bicyclic Grignard t o attain optimum transition state geometry without energetically costly distortion. The cleavage reaction of 12 t o 13 may be retarded in a similar fashion. As shown in Table I, 12 rearranges more slowly than 16b by a factor of 0.059. A correction should be made t o allow for the fact t h a t the bicyclic Grignard has a cis 2-alkyl substituent on the cyclobutane ring. Since cis- 17 cleaves to 18, with a rate about 0.3 t h a t of cyclobutylmethMgCl
17
18
CH3 19
ylmagnesium chloride,ls a corrected relative rate for the rearrangement 12 .+ 13 would be about 0.2. This decrease in rate is found despite a probable increase in thermodynamic driving force for the reaction (by release of ring strain). Unfortunately, heat of formation information is not available for the bicyclo[3.2.0]heptyl ring system, so this conclusion must be based on estimates of the strain e n e r g ~ . l ~ , ~ O Formation of the minor cycloheptenyl Grignard product 14 in significant amount warrants comment. In terms of product stability alone, this is anticipated t o be the lesser of the two products, since a secondary Grignard is about 3-5 kcal/mol less stable than a primary ne,^^-^^ and the strain energies of cyclopentene and cycloheptene are similar.21 In the simpler cleavage reaction of eq 5 , where the choice is also between formation of primary (18)and secondary (19)Grignards, the primary predominates by a factor of close to 100. In the absence of additional factors, then, a comparable ratio of ca. 100:1 might have been expected for 13:14 also. Such an additional factor may indeed exist. The most obvious model for the previously supported four-center transition state2,3 is one in which the four atoms involved are arranged in an approximately coplanar and rectangular array:
Examination of models shows t h a t such a configuration can be attained for the cleavage to the cyclopentenylethyl Grignard 13 with minimal bond bending, though probably some increase in torsional strain. However, the distortion necessary to pass through such a transition state en route t o 14 should be prohibitive. Yet, the rearrangement path t o secondary Grignard (12 .-+ 14) competes more favorably in the constrained bicycloheptyl Grignard than in the more flexible case of 17 --c 19. Furthermore, it appears t h a t the absolute rates of these two rearrangements yielding secondary Grignard are similar.28 We must therefore conclude t h a t the reaction coordinate leading to 14 more closely resembles the ideal geometry for this reaction than the one leading t o 13, and t h a t the “coplanar rectangular” model is a poor one for this transition state. Additional insight into the reaction coordinate may be drawn by viewing it in the reverse direction-as an intramolecular addition. The “coplanar-rectangular” transition state corresponds to a parallel approach of the C-Mg bond to one lobe of the C=C a orbital. In an alternative reaction coordinate, the approach might be in perpendicular fashion:
R A similar reaction coordinate has been proposed on other grounds by Eisch for the addition of triphenylaluminum to alkynes.29aThis reaction path may have the additional virtue of avoiding the unfavorable steric interactions illustrated in 20, which are present in any coplanar approach. In the present instance, we might note t h a t a substituent in t h e axial 5 position of a boat cycloheptene (21)is held in a location over the double bond, probably ideal for a “perpendicular” approach t o the addition.29b
,,
.
R /C-M R’
c-c
,*+
R’ 20
21
A reaction coordinate involving this perpendicular approach would also fit well with the notion of a a-complex intermediate in the addition. Intramolecular interaction of the metal of an organometallic with olefinic a electrons has been demonstrated for organolithium, -aluminum, and -zinc compounds,30 and r-complex intermediates are widely considered in additions of organometallics t o alkenes or alkynes.31 Such a mechanism has been suggested for the Grignard cyclization-cleavage process, though without firm experimental upp port.^-^ A possible alternative t o the logic above is t h a t the reaction proceeds by two concurrent mechanisms-a coplanar
Rearrangements of 2-Bicyclo[3.2.0]heptyl Grignard Reagents four-center cleavage t o the cyclopentenylethyl Grignard 13, and a competing radical cleavage t o t h e cycloheptenyl Grignard (eq 6). However, additional experiments render this
explanation untenable. Reduction of either 2a or 3 with trin-butyltin hydride in the absence of solvent led t o the corresponding hydrocarbon as essentially the only product. In 0.02 M hydride solution, reduction of excess 2a gave sizable amounts of 3-ethylcyclopentene, resulting from partial rearrangement of the intermediate radical 22. The absence of cycloheptene in t h e product, and of significant amounts of 3-ethylcyclopentene from reduction of 3, indicate t h a t interconversion of radicals 22 and 23 is much slower than rearrangement of 22 t o 24. Therefore, a radical mechanism cannot explain the formation of 14.
J. Org. Chem., Vol. 41, No. 7, 1976 1195 t h a t the rigidity imposed by the additional fused ring may destabilize 27 relative t o 26. The conformation with a chair cycloheptene should be badly strained-the dihedral angle between bonds in the cyclopentane ring would have t o be nearly QOO-and interconversion between boat conformations should also be quite hindered. I t may be that the added strain is severe enough t o decrease the rate of cleavage by this pathway t o a very small value. Alternatively, recyclization of 27 to 25 may occur rapidly enough t h a t the concentration of the former never reaches a significant level. T h e 6-Alkoxybicyclo[3.2.O]hept-2-yl G r i g n a r d R e agents. A study of the alkoxy-substituted Grignard reagents 31 was initiated t o investigate the possibility of concerted fragmentation processes, in which breaking of the C6-alkOXy bond occurs in the same step as CZ-Mg cleavage. n
319,R = CH, b, R = C,H,
OR
22
\
(7)
0. 23
24
3 T h e 3-Tricycl0[5.3.0.0~~~]decylG r i g n a r d (25). With the tricyclic Grignard reagent 25, rearrangement by either of the two available bond cleavages would produce a secondary Grignard reagent (eq 8). Since cleavage t o the (secon-
34 Such processes would have as driving force the conversion of a magnesium alkyl t o a magnesium alkoxide. Examination of models suggests t h a t in the transition states 35 and 36 for a concerted fragmentation, twisting and puckering of the rings allows t h e cleaving bonds t o approach an electronically favorable anti-coplanar orientation about each of the forming double bonds (though all three bonds undergoing cleavage cannot be mutually coplanar). Similarly in 37, the two eliminations are syn and anti copolanar. Coordination between the departing magnesium and alkoxy functions would provide further assistance t o cleavage, as shown in 38. However, close approach of the ether oxygen to the magnesium requires twisting away from an anti-coplanar orientation. In transition state 39,both eliminations would be anti, but the product would be trans,trans-1,4cycloheptadiene. I t is important to note t h a t 35-38 are all specific to elimination of an endo alkoxy group, but t h a t the exo epimer probably does not have a reasonable concerted pathway. Alternatively, cleavage t o yield the same products might occur in stepwise fashion, as shown in eq 10.
28
dary) cycloheptene was unexpectedly favorable in rearrangement of the bicyclic Grignard 12, it might have been anticipated t o compete even more strongly here. It was found t h a t 25 does rearrange slowly, but t h a t hydrolysis yielded only the hydrocarbons 28 and 29, corresponding t o unrearranged and the rearranged Grignard 26, respectively. There was no evidence for the presence of 30,so cleavage t o 27 must be unimportant. Examination of models suggests
On heating, Grignard reagents 31 did cleave t o yield the dienes 3-vinylcyclopentene (32) and 1,4-cycloheptadiene (33) as major and minor product, respectively. The Grignard solutions studied were mixtures of the epimers with exo and endo alkoxy groups. It was apparent, from analysis of solutions heated for different periods of time before hydrolysis, t h a t t h e epimers showed quantitative differences
1196 J. Org. Chem., Vol. 41, No. 7, 1976 in their behavior. T h e epimer with endo alkoxy decomposed substantially more rapidly. For example, although endo-31b had rearranged nearly completely after 1 h at 80°, the exo-31b in t h e same solution disappeared with a half-life of 2-3 h. Compared with t h e cleavage rearrangement of Grignard reagent 12, t h e approximate relative rates of endo- and exo-31 were about lo3 and 30, respectively. T h e products from t h e two epimers also differed. Product formed in t h e later stages of the reaction, when only exo-31 was reacting, had about 15-20% of the minor cycloheptadiene product 33. Early in t h e reaction i t was present in smaller amounts, consistent with its formation as 5-10% of the product from endo-31. A more accurate comparison between epimers was precluded by sizable amounts of disproportionation during formation of t h e Grignard. T h e increased cleavage rates of both epimers are consistent with t h e operation of an electron-withdrawing inductive effect of t h e alkoxy group, which stabilizes the intermediate structures in t h e stepwise mechanism of eq T h e greater rate with endo alkoxy could be a consequence of relief of steric repulsions present in t h e starting material. Alternatively, it may be evidence for the importance of concerted fragmentations such as 35-38, which are specific
H
36
R 35
-.
37
38
39
t o this epimer. T h e product mixture observed may be more difficult to rationalize with t h e concerted mechanism. T h e relative amounts of cleavage to t h e seven-membered ring products for both exo and endo alkoxy (15-20 and 5-lo%, respectively) are similar to t h a t found in t h e cleavage of 12 (%lo%). It requires a rather remarkable coincidence for endo-31, reacting by two or more concerted fragmentation pathways which are specific t o a single product, t o yield nearly t h e same mixture of five- and seven-membered ring products as exo-31 and 12.
Experimental Section Spectra were run on Varian Associates A-60, T-60, and HA-100 NMR spectrometers, and Beckman IR-5 and IR-8 ir spectrophotometers. Gas chromatography was on Varian Aerograph A-90-P chromatographs, using the following columns: A, 0.25 in. X 10 ft, 20% Carbowax 20M on Chromosorb P; B, 0.25 in. X 10 ft, 25% Ucon polar on firebrick; C, 0.25 in. X 10 ft, 20% poly-m-phenyl ether on Chromosorb P; D, 0.25 in. X 5 ft, Tide. Preparation and Handling of Grignard Reagents. Grignard reagents were prepared in a flask sealed to a condenser, with a side
Hill, Theissen, Cannon, Miller, Guthrie, and Chen arm above the condenser for attachment to a nitrogen source or vacuum line. The solvents used were distilled from lithium aluminum hydride under a slow stream of nitrogen and transferred by syringe. Magnesium was a sublimed grade.34 When necessary, a small crystal of 12 or a very small amount of methyl iodide, ethyl bromide, or ethylene chloride was used to help initiate the reaction. Reactions were run under a slight positive pressure of dry nitrogen. In instances where hydrolysis or disproportionation products were sufficiently volatile, the solvent and any volatile components were pumped to a cold trap under high vacuum, and replaced by fresh solvent. Samples of Grignard (ca. 0.5-1 ml) were sealed in ampules, and sometimes in an NMR tube. After heating, the ampules were opened in an inert atmosphere ''glove bag", and either hydrolyzed immediately or fitted to an adapter which allowed volatiles to be pumped to a trap before hydrolysis. Hydrocarbon products were analyzed by gas chromatography, on the assumption of equal sensitivity of isomeric hydrocarbons. Rearrangement rates were determined from a plot of log ([unrearranged]/[unrearranged] t [rearranged]) vs. time. 2-Halobicyclo[3.2.0]heptanes (2). exo-2-Bicyclo[3.2.0]heptano1 was prepared as reported by Winstein and S t a f f ~ r dbp : ~ 85-86 "c (20 mm); n25D 1.4850; NMR (CC14) 6 3.88 (d, 2, J = 3.0 Hz), 3.62 (s, 1, OH), and 1.0-3.0 ppm (m, 10). Reaction of 1.0 g (8.9 mmol) of the alcohol with thionyl chloride (1.06 g, 8.9 mmol) and pyridine (0.70 g, 8.9 mmol) yielded 0.30 g (26%)of product: bp 60 "c (20 mm); nZ4D 1.4857; NMR (CCld) 6 1.0-3.0 (m, lo), 3.89 ( 8 , broad, ca. % H), and 4.12 ppm (d, ca. % H, J = 3.7 Hz).This mixture was not studied further. To a stirred solution of the alcohol (10.0 g, 0.089 mol) and triphenylphosphine (24.5 g, 0.0935 mol) in dry dimethylformamide (50 ml) at 50-57 "C, bromine (14.5 g, 0.089 mol) was added over a period of 45 min. After an additional 1 h of heating at this temperature, the mixture was distilled under vacuum (up to 80 "C at 1 mm). The distillate was diluted with 50 ml of saturated aqueous sodium chloride and extracted with petroleum ether. The extract was washed with aqueous sodium chloride solution, distilled twice, and chromatographed on alumina to yield 4.5 g (30%)of product: bp 69" (10 mm); nZ7D1.5092; NMR (neat) 6 3.9-4.45 (m, 1, CHBr) and 1.0-3.2 ppm (m, 10). Anal. Calcd for C7H113r: C, 48.02; H, 6.33. Found: C, 48.29; H, 6.43. Gas chromatography on column A showed the presence of three components in relative amounts (in order of increasing elution time) of 12.5, 21.5, and 66%. Grignard Reagent from 2-Bromobicyclo[3.2.0]heptane. Grignard reagents were prepared from the bromide and an excess of sublimed magnesium in ether or tetrahydrofuran solution. After completion of the reaction, the solvent and any volatile materials present were pumped to a cold trap (at ca. 5 p pressure), and fresh solvent was added. Gas chromatographic examination (column B) of either the volatiles from the Grignard preparation or the product of hydrolysis of the Grignard (with cold 3 N HCl) showed two hydrocarbons. The first of these, present to the extent of 10% in the volatiles from the Grignard preparation and 13-14% in the hydrolysis product, had retention time and infrared and NMR spectra identical with those of authentic norbornane. The major component of the Grignard hydrolysis was identified as bicyclo[3.2.0]heptane on the following basis: (a) mode of formation, as major hydrolysis product from the starting halide; (b) ir spectrum, with major peaks coinciding in position with those in an authentic dilute solution spectrum kindly furnished to us by Professor W. G. Dauben; (c) NMR spectrum, which had complex absorption from 6 1.0 to 2.4 pprn and a broadened singlet at 2.75 ppm in a ratio of 10:2, but no olefinic proton absorption. The NMR spectrum of the Grignard in THF (ca. 0.8 M) had a broad, ill-defined multiplet from about 6 0.1 to -0.6 ppm, and no olefinic absorption. When the Grignard solutions were heated for a number of hours in sealed tubes in the vicinity of 100 "C, changes occurred both in the NMR spectrum and in the hydrolysis products. In the NMR, there appeared a prominent triplet at 6 -0.5 ppm ( J = 8.5 Hz, CHZMg) and olefinic absorption at 5.66 ppm (apparent AB quartet, JAB N 5.5 Hz, with probable further triplet Splitting of ca. 2 Hz). Hydrolysis yielded two new components, formed initially in a ratio of about 1O:l. The first of these was idehtified as 3-ethylcyclopentene: NMR (CC4) 6 0.91 (t, 3, J = 7 Hz, CH3), 1.0-2.8 (m, 7), and 5.63 ppm (s, 2, olefinic); ir identical with published spect r ~ mThe . ~ second ~ (minor) component was cycloheptene: NMR (CC4) 6 1.65 (m, 6), 2.13 (m, 4, allylic), and 5.72 ppm (apparent triplet, 2, J = 3.5 Hz, olefinic) and identical with an authentic sample; ir identical with published spectrum36 and authentic sam-
Rearrangements of 2-Bicyclo[3.2.0]heptyl Grignard Reagents
J.Org. Chem., Vol. 41, No. 7, 1976 1197
ple. At long heating time, the cycloheptene in the Grignard hydrolysis product decreased relative to the 3-ethylcyclopentene. At least two minor components having retention times similar to bicyclo[3.2.0]heptane were detected, but not identified. First, the NMR spectrum of the bicyclo[3.2.0]heptane in the volatiles pumped from the Grignard preparation indicated the presence of unsaturation (6 5.78 ppm, 8 ) . Integration of the spectrum would be consistent with an approximately 1:l mixture of bicyclo[3.2.0]heptane and bicyclo[3.2.0]hept-2-ene.Second, a t long heating times, the GC peak corresponding in retention time to bicyclo[3.2.0]heptane remained constant a t about 5% of the total product. Unsaturation was indicated by NMR (6 5.78 and 4.71 ppm) and infrared spectra (3060 and 802 cm-l). A saturated component (NMR singlets a t 6 0.99 and 1.20 ppm) also appeared to be present. The olefinic NMR signals were not present in the hydrolysis product of the unheated Grignard reagent. In determining the kinetics of the rearrangement, the concentration of unrearranged hydrocarbon was decreased to allow for the 5% of nonrearranging material. The total rate of disappearance of starting material was determined, and apportioned between the two rearrangement. paths according to the average ratio of ethylcyclopentene to cycloheptene. 5-Brornocycloheptene. 4-Cycloheptenol was prepared by the method of Cope? bp 68.5-70.7 "C (8 mm) [lit.6 bp 83-84 OC (11 mm)]; NMR (CC1,L) 6 1.46 (m, 2), 1.7-2.5 (m, 6), 3.73 (triplet of triplets, 1,J = 3.8,8.6 Hz, CHOH), 4.30 (s, 1, OH), 5.69 ppm (m, 2, olefinic). Phosphorus tribromide (7.08 g, 26 mmol) was added dropwise to a mixture of 4-cycloheptenol (7.9 g, 69 mmol) and pyridine (1.45 g, 18 mmol) in 15 ml of anhydrous ether cooled to 0 "C. The mixture was allowed to warm to room temperature and stand overnight. The ether was distilled a t atmospheric pressure, and then the residue was slowly heated under vacuum (-8 mm) to a maximum temperature of 105 "C'. Product distilled to a receiver. Redistillation yielded 5.8 g: bp 62-64 "C (8 mm); NMR (Cc4) 6 1.7-2.5 (m, 8), 4.39 (s, b, 1, CHBr), 5.76 ppm (m, 2, olefinic). Gas chromatography (column B) showed the presence of a minor component (ca. 10%) eluted just ahead of the major bromide. Anal. Calcd for C7HI1Br: C, 48.02; H, 6.33. Found: C, 47.85; H, 6.28. Grignard Reagent from 5-Bromocycloheptene. A Grignard reagent was prepared from 5-bromocycloheptene and magnesium in dried ether. Hydrolysis of the Grignard reagent yielded a mixture of hydrocarbons consisting principally of cycloheptene (86%) and norbornane (9%), as shown by GC analysis (column B). Two minor components of longer retention time, both unsaturated, were present. The solvent pumped from the Grignard preparation had the same components, with the norbornane and cycloheptene comprising about 50% of the total. The ir and NMR spectra of the major longer retention time component were identical with those of 1,4-cycloheptadiene (see below). An NMR spectrum of the Grignard solution in ether showed olefinic absorption at 6 5.7 ppm as an unsymmetrical complex multiplet, about 15 Hz wide. The CHMg resonance was obscured by the solvent side bands in ether, but appeared a t 6 0.5 ppm in THF solution. It was an apparent pentuplet, with J = 5.5 Hz. A weak broad doublet absorption a t 6 -0.4 ppm was not identified. The Grignard solution in ether was heated for 92 h at 130 "C. Volatiles were removed under vacuum and the residual Grignard was hydrolyzed. As before, the product contained cycloheptene and norbornane, but the major component was identified as 3-ethylcyclopentene. About 72% of the cycloheptenyl component had undergone rearrangement. The volatiles pumped from the Grignard before hydrolysis contained the same components as previously, with cycloheptene comprising the major portion. The norbornane appeared to have remained as a constant 8-10% of the total products. A sample heated for 120 h a t 110 OC showed a smaller fraction of rearrangement (6-7%). The NMR spectrum of the heated Grignard reagent showed additional rather ill-defined olefinic absorption at 6 5.45-5.65, and a sharp singlet a t 5.33 ppm attributed to ethylene formed by attack of Grignard on the ether solvent. A high-field CHzMg triplet appeared at -0.50 ppm, with J = 8.5 Hz. 6-Alkoxybicyclo[3.2.O]heptan-2-ones (4a and 4b). An excess of methyl vinyl ether (3-4 ml) was condensed into Pyrex ampules of 1-cm diameter. 2-Cyclopenten-1-one (2 g) was added to each, and the ampules were sealed under vacuum. The ampules were exposed for 24 h to the output of a Hanovia 450-W mercury arc lamp, while immersed in a bath of water at about 10 "C. After irradiation, the ampules were opened and combined, and the mixture dis-
tilled to yield 71% of 4a: bp 216, 110-112 "C (20 mm); ir (Cc4) 1748, 1124 cm-I; NMR (cc14)6 4.1-3.4 (m, 1, CHO), 3.17 (s, 3, OCH3), and 3.1-1.4 ppm (m, 8). Gas chromatography (columns C and D) showed that the product was a mixture of two incompletely separated components present in roughly equal amount. Anal. Calcd for CSH1202: C, 68.57, H, 8.57. Found: C, 68.28; H, 8.49. A similar reaction of cyclopentenone (25 g, 0.30 ml) and ethyl vinyl ether (50 ml) yielded a mixture of 4b: bp 62-65 "C (1 mm); ir (CC14)1742, 1120 cm-l; NMR (Cc4) 6 4.25-3.6 (m, 1, CHO), 3.32 Jq, 2, J = 7 Hz, CHzO), 3.1-1.35 (m, 8), 1.4 ppm (t, 3, J = 7 Hz, CH3). Gas chromatography (column C) again showed a mixture of two partially separated components in approximately a 1:l ratio. Anal. Calcd for CgH1402: C, 70.13; H, 9.09. Found: C, 70.02; H, 8.83. 6-Alkoxybicyclo[3.2.O]heptanes (34a and 34b). A mixture of methoxy ketone 4a (2.52 g, 0.018 mol), 85% hydrazine hydrate (20 ml), and diethylene glycol (20 ml) was heated at reflux for 45 min. The mixture was cooled, potassium hydroxide (6.7 g, 0.12 mol) was added, and the mixture was distilled slowly up to a pot temperature of 190 "C. The distillate was extracted with pentane, dried (NazSOd), and distilled to yield 1.15 g (45%) of 34a, ir (CC14) 1123 cm-l. The product was a mixture of two components present in equal amount, which were separated by preparative GC (column C, in order of increasing retention time): (1) endo methoxy, NMR (CC14) 6 4.1-3.3 (m, 1, CHO), 3.10 ( 6 , 3, OCH3), and 3.0-1.25 (m, 10); and (2) exo methoxy, NMR (CClI) 6 3.6-3.15 (m, 1,CHO), 3.10 (s, 3, OCH3), 2.6 (s, b, 2, bridgehead), and 2.35-1.15 ppm (m, 8). Anal. Calcd for C8HI4O: C, 76.14, H, 11.18. Found: C, 76.23; H, 10.95. A similar mixture was produced by reaction of bromide 6a (1 g) with an equimolar quantity of tri-n-butyltin hydride in refluxing benzene. Reduction of 6a (1 g) by addition as a solution in 2 ml of ethanol to 0.25 g of molten sodium yielded a mixture of products shown by NMR to contain some unsaturation. On GC column D, the product was separated into two portions. The first of these 1640 (C=C) and 1078 cm-'; NMR 6 5.7 (18%) had ir (b, 2, =CH), 3.75-3.25 (m, 1, CHO), 3.15 (s, 3.0 CH3), and 3.1-1.9 ppm (m, 6), and was tentatively identified as endo-6-methoxybicyclo[3.2.0]hept-2-ene. The second appeared to contain mostly the expected saturated reduction product, along with about 15-20% of an olefinic component. Wolff-Kishner reduction of the ethoxy ketone 4b was carried out in similar fashion: ir (CCl4) 1133 and 1108 cm-I. The components were separated by preparative gas chromatography on column C: (1) endo ethoxy, NMR (cc14)6 3.82 (9, b, 1, J = 7.5 Hz, CHO), 3.27 (q, 2, J = 6.8 Hz, CH20), 3.0-1.3 (m, lo), and 1.11ppm (t, 3, J = 6.8 Hz, CH3); (2) exo ethoxy, NMR (CCld) 6 3.3 (9, superimposed on an unresolved multiplet, 3, JqUBrtet = 6.8 Hz, CHO and CHzO), 2.6 (s, b, 2, bridgehead), 2.4-1.3 (m, 8), 1.11 (t, 3, J = 6.8, '333).
6-Alkoxybicyclo[3.2.O]-heptan-endo-2-ols (5a and 5b). A suspension of methoxy ketone 4a (10.9 g, 0.072 mol) ih 50 ml of 2 N sodium hydroxide was added in portions to a solution of sodium borohydride (2.0 g, 0.072 mol) in 15 ml of 2 N sodium hydroxide. The mixture was stirred and maintained at 35-40 OC by cooling during the addition, then heated gradually to 90 "C over 2 h and maintained at that temperature for 1.5 h. Anhydrous potassium carbonate (9.6 g) was added, the organic layer was separated, and the aqueous phase was extracted with ether. The organic solutions were dried (NazS04) and distilled to yield 7.5 g (68.5%) of 5a: bp 127-130° (20 mm); ir (cc14)3457,1207,1120, and 1072 cm-'. The product was found by GC to consist of two components which were preparatively separated (column D) for spectroscopic characterization: (1) endo methoxy, NMR 6 4.1-3.4 (m, 3, OH and CHO), 3.13 (s, 3, OCH3), and 2.9-1.2 ppm (m, 8); (2) exo methoxy, NMR (cc14) 6 4.2-3.7 (m, 2, OH and CzHO), 3.25 (m, 1, CsHO), 3.13 (s, 3, OCHd, 2.5 (s, b, 2, bridgehead), and 2.5-1.3 (m, 6). Anal. Calcd for CaH1402: C, 67.60; H, 9.86. Found: C, 67.57; H, 9.88. The ethoxy ketone 4b was reduced to 5b by addition of sodium borohydride (2.5 g, 0.066 mol) in small portions to a solution of ketone (20.9 g, 0.135 mol) in 100 ml of absolute ethanol. The mixture was refluxed for 3 h, 100 g of 25% aqueous sodium hydroxide was added, reflux was continued for 1 h, and 100 ml of water was added. The product was extracted into ether, washed, dried, and distilled to yield 18.7 g (90%): bp 77-81 "C (1 mm); ir (CC14) 3454, 1192, 1120, and 1070 cm-l. The mixture had NMR absorption for the ethoxy group at 3.28 and 1.13 ppm ( J = 7 Hz) and multiplet absorption in the region 4.4-3.5 ppm attributable to the CHO pro-
1198 J . Org. Chem., Vol. 41, No. 7, 1976
Hill, Theissen, Cannon, Miller, Guthrie, and Chen
tons. Partial resolution could be obtained by gas chromatography (column C), but spectra of the separated fractions were relatively uninformative. Anal. Calcd for C9H1602: C, 69.20; H, 10.32. Found: C, 69.15; H, 10.49. exo-2-Bromo-6-alkoxybicyclo[3.2.O]heptanes (sa and 6b). Bromine (7.2 g, 0.045 mol) was added dropwise over a period of 30 min to a stirred solution of 5a (6.4 g, 0.045 mol) and triphenylphosphine (11.8 g, 0.045 mol) in 80 ml of dry dimethylformamide, maintained at 55 "C. After 2.5 h of additional heating at 55 "C, the reaction mixture was distilled, up to pot temperature of 90 "C (1 mm). The distillate was extracted with 30-60 "C petroleum ether. The extract was washed (aqueous NaCl), dried (CaClZ), and distilled to give a 61% yield of 6a: bp 69-70 "C (1mm); ir (CC14) 1207, 1120, and 1072 cm-l. Analysis by gas chromatography (columns C and D) showed the presence of two components, which were preparatively separated for spectroscopic characterization: (1) endo methoxy, NMR (CC4) 6 4.25 (d, 1, J 3 Hz, CHBr), 4.0-3.5 (m, 1, CHO), 3.14 ( 8 , 3, OCHs), and 3.1-1.0 ppm (m, 8); (2) exo methoxy, NMR (CCl4) 6 4.25 (d, 1, J 3 Hz, CHBr), 3.3 (m, 1, CHO), 3.14 (s, 3, OCHs), and 3.1-1.0 ppm (m, 8). Anal. Calcd for CgHlsBrO: C, 46.85; H, 6.34. Found: C, 46.88; H, 6.45. In analogous fashion, 5b was converted to 6b in 53% yield: bp 90" (1 mm); ir (CC4) 2999, 1291, 1229, 1199, 1133, 1119, 898, 734, and 711 cm-l. The NMR spectrum had a total of four hydrogens at low field (4.4-3.0 ppm), including a one-proton doublet a t 4.20 ppm ( J 3 Hz, CHBr), a two-proton quartet at 3.27 ppm ( J = 7 Hz, CHzO), and additional absorption totaling about one hydrogen, partially obscured by the ethoxy quartet, and partially appearing as an apparent quartet at 3.84 ppm ( J 7 Hz). Along with complex absorption from 3.0 to 1.3 ppm, the methyl triplet appeared at 1.12 ppm ( J = 7 Hz). Gas chromatography (column C) indicated a mixture of two components, present in approximately equal amounts. Grignard Reagents from 6a and 6b. Grignard reagents were prepared from 0.5-2-g samples of 6a or 6b and an excess of sublimed magnesium in ether, following the general procedure. Concentrations were determined by acid titration and, in some cases, by complexiometric titration for magnesium. In either case, hydrolysis yielded a similar mixture which was partially resolved by gas chromatography (columns C and D). The major GC fraction was a broadened peak which included both exo and endo isomers of the appropriate 6-alkoxybicyclo[3.2.O]heptane,along with small amounts of an unsaturated component tentatively assigned as the exo-6-alkoxybicyclo[3.2.0] hept-2-ene. The probable endo isomer of the latter was eluted shortly before the major peak. This portion of the product closely resembled the product of sodium-ethanol reduction of 6a (see above). The unsaturated components, which are presumed to have originated in a side reaction during Grignard reagent formation, comprised about 25-30% of the total product. On heating for a period of hours in the vicinity of 80 "C two new components appeared, which increased with time at the expense of the major fraction. These were identified by comparison of ir and NMR spectra and GC retention times with authentic samples of 3-vinylcyclopentene (32) and 1,4-cycloheptadiene (33) (in order of retention time). In some cases, the Grignard reagent appeared to have significant amounts of the former before heating. They were formed in a ratio of 5 to 1O:l. In addition, both Grignard reagents contained a small amount (